For multicluster support, we will be going with a solution that has a centralized Jaeger instance that receives tracing data from multiple clusters. Communication between the clusters will be accomplished using opentelemetry collectors.

Tasks

• determine the workflow for how this will work
• generate an architecture diagram
• determine security and how this will work (end to end: how a customer will setup and configure things)
• determine how this could work with observatorium. We don't have a separate experience between using observatorium and using a more standalone Jaeger.
• determine how updates will work with this.
• Migrating from existing separate clusters to having traces being store centrally.
• Determine how this will work with OSSM
• Make sure that users can determine which data is coming from each cluster

Out of scope

• RBAC
• caching of traces when networks are down
• scalability / performance of a single Jaeger instance
• changes to Jaeger or elasticsearch/storage. The work should be focused on opentelemetry collectors

Deliverables

• doc explaining things
• Jiras
• agreement with the engineering team and PM